[ On Saturday, March 15, 2003 at 22:53:54 (+0000), Giuliano Gavazzi wrote: ]
> Subject: Re: [Exim] Reaction to rude 554 greeting
>
> At 20:29 +0000 2003/03/15, Philip Hazel wrote:
> [...]
> > Right. So if an address is routed to such a host, something is screwed
> > up (unless this is deliberate). Bouncing the message causes someone to
> > pay attention.
>
> This makes sense, but it does not help in making the delivery.
The point is that the delivery MUST NOT be attempted any further!
The point is to handle e-mail reliably as possible, not to make guesses
as to where it might be intended to go. Human intervention is always
required in these situations!
> Perhaps one could generate a warning message (back to the sender) and
> attempt to deliver to another MX.
No, absolutely not. The message _MUST_ be bounced immediately! That's
what a 5xx response code _always_ means!
> Contrary to what Greg said in
> another message, it seems that retries can be at different intervals:
>
> The sender MUST delay retrying a particular destination after one
> attempt has failed. In general, the retry interval SHOULD be at
> least 30 minutes; however, more sophisticated and variable strategies
> will be beneficial when the SMTP client can determine the reason for
> non-delivery.
You are missing how this fits into the bigger picture. The above
paragraph is not in any way contrary to what I said. Indeed if you're
referring to what I think you're referring to then I was paraphrasing
from that very paragraph!
> in the present case, since we clearly hit a non-smtp server, we could
> try one other MX immediately.
No, the target server is _NOT_ "clearly a non-smtp server"!
Whether or not another MX may be tried for the delivery of another
message to the same destination depends _entirely_ on whether another MX
might be chosen by the well defined MX routing algorithm. I.e. another
MX may be used for the same destination _only_ if that other MX host has
equivalent priority to the one which failed.
Yes this does mean that idiots without a clue can create a situation
where only some of the e-mail destined to their domain will ever be
delivered and the rest will always be bounced. That's their problem
though, not the sender's problem.
> although there is one case when 5xx should be treated as 4xx (552 as
> 452),
Please do not confuse matters any more than they already are by taking
yet another example so completely out of context! The treatment of 552
as 452 is _only_ allowed for the RCPT command, and even then the advice
of RFC 2821 is highly quesionable in this regard.
--
Greg A. Woods
+1 416 218-0098; <g.a.woods@???>; <woods@???>
Planix, Inc. <woods@???>; VE3TCP; Secrets of the Weird <woods@???>
